As vision algorithms are developed and applied to unmanned aerial vehicle (UAV) platforms, intelligent tracking has achieved desirable tracking effects. How to implement an entire unmanned process from tracking to landing is an important direction for making unmanned aerial vehicles more intelligent. In an existing unmanned aerial vehicle technology, a function of returning home by pressing one key can enable an aerial vehicle to autonomously fly to a specified place according to a one-off instruction of a user, and flight and landing processes no longer require involvement of a user. However, this technology is limited to returning to and landing on a fixed location.
Currently, an unmanned aerial vehicle in the prior art is already capable of precisely landing on a special returning location that moves continuously such as a car or a ship. However, this technology requires that an external measurement apparatus (for example, a GPS) obtains motion information of a moving target. Besides, a method of fusion image feature matching so that obtaining a precise landing location requires inputting image information of a target in advance. If a target is not equipped with a measurement apparatus or a returning target changes in a flight process, this method becomes invalid.